FORMAL AND FUNCTIONAL APPROACHES TO THE STUDY OF
LANGUAGE-2
A module for a course on ‘Language Studies’ for the UGC e-PG Pathshala
Programme for the Creation of e-Contents for a Post-Graduate Course in Linguistics
-Pramod Pandey, JNU, New Delhi
Objectives:
To familiarize students with the two distinct approaches to language study and
to show that there is need to integrate them.
Contents:
1. Introduction
2 The Concepts ‘Formal’ and ‘Functional’
3 Linguistic Approaches
4. Complementariness of the formalist and functionalist approaches
5. Distinctive Feature Theory as an exemplar case of the integration of formal
and functional approaches to language study.
6. Some typical areas dominated by the formal and functional approaches
7. Summary
1. Introduction
In this module, we introduce you to two main approaches to the study of
language, namely, formal and functional approaches. The differences between
them are not well defined from the start. Rather they have grown with time, as
linguists came to assume the ‘primary’ or ‘main’ nature of language being the
formal aspects of linguistic structure or the functional aspects dealing with
communication. Linguistic theories since Structural Linguistics can be said to
cluster around these two assumptions regarding the essential or primary
properties of language. In what follows, we take a look at some of the main
aspects of these two approaches. An understanding of the main concerns of the
two approaches will help you to choose a theory or theories that best suit your
interest in the study of language.
2
The Concepts ‘Formal’ and ‘Functional’
The terms formal and functional are used for more than one aspect of linguistic
study in current literature in modern linguistics. Three main aspects are
prominent prominent. These are (i) the nature of linguistic knowledge, (ii) the
goals of linguistic analysis and (iii) types of linguistic explanation. The
differences between the two approaches are relatively more clearly drawn with
regard to these aspects than others. We will look at the other aspects, too, in a
following section.
Towards the end of the module, we will see that there is need to see the
standpoints of the two approaches as complementary rather than contending as
is sometimes found to guide linguistic studies in the two approaches.
3
Linguistic Approaches
As discussed in the previous module, current approaches to the study of
language can be broadly classified into two types: Formal and Functional (see
e.g. Newmeyer 1998). An exemplar theory of the formal approach to the study of
language is Generative Linguistics, proposed and promulgated by Chomsky in
various publications (e.g. Chomsky 1957, Chomsky, 1965, Chomsky 1982,
Chomsky 1998). Well-known theories of the functional approach are Functional
Sentence Perspective (e.g. Firbas 1992), Systemic Functional Grammar (e.g.
Halliday 1985) and Functional Grammar (e.g. Givon 1979, Dik 1989, Foley and
van Valin 1985). Comparative studies of the two approaches are available in
Darnell et al. (1998), and Newmeyer (1998). Individual studies of the main
concerns of the formal approach are found in Chomsky (1959, 1990) and Cook
(2007), amongst a plethora of publications. A good analytical review of
functional approaches is available in Nichols (1984).
A comparative understanding of the two approaches is of interest not only to
linguistics but also to epistemology in general, as they represent two different
conceptions of linguistic knowledge. They have analogues in other areas, such as
rationalism and empiricism in philosophy and cognitive psychology, capitalism
and communism in economic thought, and individualism and socialism in social
action. The basic premises and methods of the formalist and functionalist
approaches in linguistics are often stated, largely assumed, although not always
agreed upon (see e.g. Haspelmath 2000). The following may be stated as the
main theoretical issues on which the two approaches arguably differ (see also
Pandey 1999).
(1)
THEORETICAL GOALS
NATURE OF LINGUISTIC KNOWLEDGE
RESEARCH METHOD
RELEVANT DATA
NATURE OF EXPLANATION
NATURE OF ARGUMENTATION
THEORETICAL CATEGORIES (CONCEPTS AND DEVICES)
We discuss these issues in simple words briefly below.
3.1
THEORETICAL GOALS
The central concern of the formal approach is the explanation of linguistic
knowledge. For example, consider the following sentences
(2)
a.
b.
The cycle was found by the police.
The cycle was found by the road.
(3)
a.
b.
The police found the cycle.
The cycle was found by the police.
A speaker-hearer of English understands the two sentences in (2) differently,
although they are identical on the surface. In (2a) the police found the cycle, and
in (2b), someone found the cycle by the road. In (3), the two sentences are
understood as having the same meaning. Opposite of (2), the sentences are
different in their structures on the surface. The goal of a formal theory of
language is to investigate the knowledge that underlies the similarities and
differences in the understanding and production of these sentences. This
knowledge is considered to be the native speaker-hearer’s COMPETENCE.
The central concern of the functional approach is the explanation of linguistic
texts. For instance, in the functional approach, given the sentences in (2) and (3)
above, a relevant question would be, what functions do they perform? What
determines the choice of active and passive sentences? What is the relation
between the verb, on the one hand, and the Subject and the Object of the
sentences, on the other?
There are various other concerns with the text in functional approaches. For
instance, what devices are used in answering questions in different ways- formal,
polite, intimate, impolite etc. (Hymes 1985)? What devices of cohesion and
coherence are used in holding discourse among the speakers of a language
(Halliday & Hasan 1976)? The nature of linguistic knowledge assumed here has
been called COMMUNICATIVE COMPETENCE.
3.2
NATURE OF LINGUISTIC KNOWLEDGE
In the post generative grammar era, that is, post-Noam Chomsky’s Syntactic
Structures (1957), formal theories came to consider the biologically endowed
knowledge of language to be the main object of explanation (Chomsky 1986).
Formal theories assume that a human being with normal faculties is born with
the ability to know language. The human mind/brain comes equipped with the
essential properties of grammar, also known as Universal Grammar (Cook 2007).
It is the internalized knowledge of language that a formal linguist aims to
investigate. The focus is not the utterances on the surface but the knowledge
underlying the utterances. With regard to the sentences in (2) and (3) above, the
main concern in the formal approach would be to explore the cognitive basis of
the relation between them. In so doing, a formal approach aims at explaining a
speaker-hearer’s intuition guiding his linguistic knowledge. The intuition is
abstract and is related to the surface by means of certain cognitive devices. The
nature of linguistic intuition regarding utterances and the devices that connect
intuition with the surface utterances is of central concern in formal theories of
language. You will find these elaborated in the courses on Phonology,
Morphology, Syntax and Formal Semantics in many of the papers. It is the far-
reaching results of the formal investigations of language, along the lines argued
for by Noam Chomsky, that the discipline of Cognitive Science has come about,
drawing researchers from various fields such as linguistics, information science,
cognitive psychology and applied linguistics.
In the functional approach, from the time of the Prague Linguistic Circle of the
1920’s, linguistic structure was of interest not in itself but as an instrument of
communication. Sentence structures were studied from the point of view of the
functions they performed. In so doing it is not the linguistic knowledge
underlying the sentences but the communicative differences and similarities of
spoken utterances and written texts that is the goal of study. We can use the
term ‘text’ to refer to both spoken utterances and written texts.
There are many other aspects of linguistic knowledge that are of especial
interest to functional linguists. For instance, in what situations do speakers of a
language switch between varieties of a language, such as a colloquial variety, an
informal variety, etc.? What are the communicative concerns in the use of
different varieties in different linguistic situations? Much work in the field of
sociolinguistics (e.g. Labov 1968) is concerned with issues such as these. You
find them discussed, for example, in the modules of Paper 4 on Pragmatics and
Discourse Analysis and Paper 9(B) on sociolinguistics.
3.3
RESEARCH METHOD: Natural science versus social science
The methods of research employed in the two approaches are different. Whereas
the formalist theories use the methods of natural science, functional theories use
the methods of social science. Formal theories are based on the formulation of
hypotheses, choosing between hypotheses, search for evidence in support of a
hypothesis, making predictions based on the evidence in favour of the
hypotheses. The predictions involve formal generalizations in terms of
unambiguous and verifiable statements. The statements must be refutable in the
sense of Popper (1963).
The methods of research employed in the functional approaches involve the
research methods of social sciences, which take into account the variables
involved in linguistic behavior- speakers, contexts, relation between
interlocutors, etc.. Studies in the functionalist approaches also address questions
that are of concern in social sciences, such as identities, speech acts (linked to
multiple issues such as the ethnography of speaking, social network, politeness
theory and stylistics)
The data collected from a set of speakers of a community are then analysed using
qualitative and quantitative methods.
3.4
RELEVANT DATA: Intuition versus behaviour
The data considered relevant in formal approaches are the speaker-hearer’s
intuitions. This involves questions such as the following. Do speakers consider
two utterances as having the same meaning or different? Given a set of
utterances, which ones do the speakers consider to be grammatical or
ungrammatical? What is the nature of language processing with regard to certain
phenomena. Much of the data produced in support of formal investigations also
involves errors, negative data that speakers consider ungrammatical etc. The
study of linguistic behavior is of no concern to formal theories.
The relevant data for functional approaches are ones that relate to linguistic
behavior and the use of language in different contexts. The relation of context to
data is of main concern in functional approaches (Halliday 1985)
3.5
NATURE OF EXPLANATION: Deductive-Nomological versus teleological and
statistical.
There does, however, appear to be some sort of a demarcation, not strictly a
fence, between the two types. What is the line of demarcation between formal
and functional explanations, if any?
a.
The basic difference between the two types of explanation can be
attributed to the requirement for formal explanations to be stated, using a
common conceptual vocabulary, so that they are falsifiable, but non-falsified. The
distinction between the italicized expressions is crucial. A formal statement may
include any type of concepts, formal or functional, but the statement should be
refutable. If refuted, however, it must presumably be revised. Many notions and
principle in generative grammar have come to be revised in the face of data that
are not amenable to analysis. Many rules called transformational rules that were
proposed in generative grammar were gradually modified and finally dropped
from the theory as they were falsified. Falsifiability is a strict criterion for a
formal explanation, but not for a functional explanation. Many functional
principles proposed in linguistics are within the context of one of the many
functional theories, which are not a part of a universal theory. Linguistic
principles in them are proposed such that they are not amenable to falsification.
And even when they are falsified, the theory need not modify them.
As an example, let us consider, as illustrations of this point of difference,
treatments of a phenomenon in the two approaches. One such topic is the Noun
Phrase in English. A formalist treatment of the Noun Phrase is found in Abney
(1987), a functionalist treatment of it is found in Givon (1993). Abbney’s
treatment takes a comprehensive stock of all the possible and impossible types
of NPs and how they are predicted by the revised X-bar theory of phrase
structures. In contrast, Givon’s account of NPs provides a description of the
structures involved in terms of his own version of Functional Grammar.
But note: the depth and range of formal accounts is offset by their avowed
exclusion of communicative use. Aspects of structure which are dependent on
use find a more revealing treatment in functional accounts. Thus an examination
of the aspects of reference and definiteness of NPs, which is bound in discourse
(such as the knowledge of the world, speaker’s intuition, etc.) is found to unfurl a
deeper range of facts about the knowledge of NPs in Givon (1993) that a reader
of Abbney must remain ignorant of.
b.
A second requirement of a formal explanation, based on the work on
formal properties of cognition (e.g. Pylyshyn 1984, Bromberger and Halle 1989),
and related to the one discussed above, is that it concern the computational
faculty of language rather than the adaptive faculty, and, by implication, to
invariant rather than emergent phenomena. The latter, being bound to contexts,
may be unpredictable and thus stand falsified. According to this criterion
explanations of sociolinguistic variations cannot be treated as formal, since they
address the adaptive rather than the computational linguistic faculty. On the
same grounds, the explanation of gradient phonetic phenomena referring to
adaptive or emergent speech using continuous mathematics (e.g. Pierrehumbert
et al. 2000) may not be considered formal, unless shown to be related to the
computational faculty of speech.
But note: it is worth stating (and for students of linguistics to note) that for an
explanation to be formal or functional does not affect its validity for scientific
research, contrary to the assumptions of hard core formalists (e.g. Hale and Reiss
2000, van der Hulst 2000). Whether or not one of them is more valuable than the
other in linguistic theorizing is controversial at this stage.
3.6
NATURE OF ARGUMENTATION: Deductive and based on negative evidence
versus inductive and textual.
Like theories of natural science, a formal linguistic theory begins with deductive
generalizations, and seeks data to refute the generalizations. For instance,
Chomsky claimed that one of the main properties of Universal Grammar is
recursion or recursivity. A general sense in which recursion as a property of
language is understood is the property of taking the output of an input as the
next input. A constituent is recursive if it embeds “a constituent in a constituent
of the same type”. (Pinker & Jackendoff 2005: 211). A sentence of this kind that
you may be familiar with is This is the dog [[that chased the cat] that ate the rat]
that dug the hole]]]…
In this sentence a ‘that clause’ is embedded in another that clause. It is not
difficult to see that recursive structures give langue the property of an infinite
structure.
As a generalization, it calls for refutation. There hasn’t been argument any put
forth against the property of recursion in human language. The attestation of a
linguistic generalization is sought through negative evidence. For instance,
suppose the formal structure of a sentence is proposed to be
(2)
i.
ii.
S
VP
à NP + VP
à V + NP
Sentence such as the following will attest the correctness of the rules:
(3)
John thanked the cop.
The young man wearing a blue hat drove the car.
However, when a sentence such as the following is encountered
(4)
The young man wept.
The rule VP à V + NP must be modified as VP à V (+ NP). Sentences such as The
young man wept for a long time, further motivate a modification in the rule S à
NP + VP. (How? Argue.)
Contrary to a formal method of argumentation, the method followed in
functional theories is that of looking at the data and then arriving at a
generalization that is stated in such as way that it is not refuted. For instance, a
survey of Eastern Hindi dialect may show that a majority its educated speakers
lack a post-alveolar sibilant /ʃ/ in their speech. They only have the dental/
alveolar /s/. However, it is likely that some educated speakers of the dialect do
have both the sibilants. In the face of this data, the observation regarding the
absence of /ʃ/ in educated eastern Hindi variety need not be shown to be wrong.
It is assumed that such a generalization has counterevidence.
3.7
THEORETICAL CATEGORIES (CONCEPTS AND DEVICES): Formal categories versus
functional categories.
Categorization is one of the essential features of human cognition and thus of
language. In linguistic descriptions, a distinction is made between formal
categories and functional or semantic categories. Formal categories defined in
terms of the formal structure of linguistic units and/or their distribution.
For
instance, the categories S, NP, VP, DET, ADVP, PP. Each one of them is defined in
terms of their structure. An S is a category that consists of NP + VP. A NOUN, and
can be defined formally as a constitute that takes NUMBER and CASE affixes.
Formally defined categories were found to be better suited for scientific
discussions as were consistent verifiable. Semantic categories on the other hand
could not be defined in a consistent manner. For instance, the semantic
definition of a noun in traditional grammar as ‘the name of a person, place or
thing’ cannot be found to apply to words such as ‘darkness or ‘joy’. It is on these
grounds that at the levels of morphology and syntax semantic/ functional
categories were found to be problematic. Thus the use of the terms Subject and
Object defined in terms of the widely held semantic definition of ‘actor’ or ‘doer’
and ‘goal’, respectively, could not be found to be generally applicable to all
Subjects and Objects, as Chomsky (1957) pointed out. Thus lunch in the sentence
The lunch pleased the aardvark very much could not be considered an actor or
doer in any of the senses of these terms. The general use of formal categories
was practised in many areas of grammar and its use in allied disciplines such as
language processing and machine translation. A whole field of semantics called
Formal Semantics was developled along the lines of formal grammar. You have
Paper 9(A) devoted to the field.
However, in-depth analyses of grammatical properties in specific languages, such
as English, and cross-linguistic descriptions of languages, especially in language
typology showed that a restriction to either of the two types of grammatical
categories could not be used to account for linguistic phenomena. Most theories
of grammar today use a combination of formal and semantic categories. Thus the
concepts of Subject, Predicate, Topic, Focus, etc. are in use in generative
grammar today, which originally rejected the use of semantic categories in its
theoretical framework.
4. Complementariness of the formalist and functionalist approaches
Notwithstanding the professed and acknowledged differences between the two
approaches, there has been a growing realization of their mutual
complementariness, as pointed out at many places in the preceding section. The
complementariness of the two approaches is apparent in at least three ways.
4.1.
Differences valid for different domains
Both formal and functional approaches have been found to be valid in different
domains. Thus the account of the grammatical principles of syntactic processes
has been found to be immensely deepened by generative linguistic accounts,
while at the same time certain syntactic phenomena, such as word order
universals (e.g. Tomlin 1986), have been found to be best explainable in
functionalist terms (e.g. Comrie 1981).
4.2
Language as an Object
Some philosophers, such as Pateman (1989) and Gellner (1998), have shown
that language, as an object, is neither only a natural kind (i.e., biologically
determined) nor a social kind (i.e. socially constructed), but both a natural and
social kind. As such, language is amenable to both formal and functional
approaches to research and explanation.
4.3 Growing rise in the mutual use of concepts between the two approaches
Attempts to unify the two approaches have been on the rise on both sides, in the
formalist work of North American linguists (see e.g. Newmeyer 2001, for a
review), and in the functionalist work of European linguists (e.g. Dik 1989 and
Sperber and Wilson 1995). This trend is remarkably noticeable in the core
formalist work, where construction-specific rules have been replaced by general
principles. As it turns out, many of these principles use functional concepts, such
as the universal ‘Economy Principle’2 and the principal of ‘Greed’3 in Minimalist
Program (Chomsky 1995), determining the constituent structure of sentences
and their movement. Within formal phonology, the rise of Optimality Theory (e.g.
Prince and Smolensky 1993) represents a rather extreme step in the direction of
the use of universal functional principles alone.
5. Distinctive Feature Theory as an exemplar case of the integration of
formal and functional approaches to language study.
It is now considered an axiomatic fact about speech sound segments that they
are made up of distinctive features, which commonly have binary (+/-) values, as
shown for a five-vowel system below:
Figure-1: Distinctive feature specifications for a five vowel system
The theory of distinctive features has had a continuous history from the time of
Trubebtzkoy (1939) and Jakobson (1968). It has undergone various revisions
(see e.g. Jakobson, Fant & Halle 1952, Chomsky & Halle 1968, Clements 1985,
McCarthy 1988). The crucial point to note here is that the development of the
theory presents itself as an exemplar case of the integration of form and function
in linguistic knowledge. Let us look at some important points of the theory.
The concept “distinctive feature” owes its origin to the functional orientation of
Prague School. Trubebtzkoy (1939), and perhaps Jakobson (1968 [1932]) before
him, proposed it as a phonetic property constituting a unit of distinctive contrast.
The phonetic property combines with other distinctive phonetic properties to
yield “the inner order or structure of the phonemic inventory as a system of
distinctive oppositions” (p. 71). For example, the feature “voice” is in “privative”
contrast in /b/: /p/; the feature is present in one and absent in the other
segment.
The development of the theory of distinctive features has taken place in the
climate of formal linguistics, in the main, Generative Phonology (Chomsky &
Halle 1968) and its sub-theories. The significance of the notion of distinctive
feature in the phonological analysis of languages was found to extend itself to
other aspects of phonological theory, such as a universal theory of phonological
oppositions (e.g Jakobson & Halle 1956, Chomsky & Halle 1968), a universal
theory of possible speech sounds (e.g. Chomsky & Halle 1968), and the
characterization of natural classes of speech sounds (e.g. Chomsky & Halle 1968,
Clements 1985). For classic arguments in favour of the formal view presented
above, see Halle (1962). Much research in the distinctive feature theory can be
found to have been devoted to the question of the exact number and definitions
of such a finite set.
The current view of the organization of the distinctive features can be
represented in the following diagram (see Broe 1992):
Figure-2: hierarchical organization of distinctive features
The non-linear organization of the features is found to account for all the
segment types in world languages and all possible processes of change due to
assimilation (i.e. adjacent occurrence). The hierarchical organization of the
features in Figure-2 also implies their autonomous representation, along the
lines of the autonomous representation of other phonological units, as claimed
by Autosegmental Phonology (see Goldsmith 1976, McCarthy 1979). Figure-2
thus would not have been envisaged without the advent of the theory of
Autosegmental Phonology. Both the theories support each other as theories of
phonological representation.
The development of the notion of distinctive features has found formal accounts
in the theory of Underspecification. The theory of underspecification deals with
the issue of simplifying the grammar by avoiding the specification of all the
features at the underlying representation and filling them out by Redundancy
rules at later stages in the phonological derivation (e.g. Archangeli 1988, Steriade
1987, Clements 1988).
In addition to the formal issues of the representation of distinctive features in
terms of a minimal set, the notion has also been invoked to account for the
structure of phonological inventories (see Module 32 in Course 13 ‘Linguistic
Typology and Language Universals’ for a detailed discussion.
6. Some typical areas dominated by the formal and functional approaches
Notwithstanding the significance of the integration of the formal and functional
approaches, it is useful to see which areas are subject to alone of the approaches
predominantly.
Thus, the following areas of study are predominantly functional:
Computational Linguistics: Machine Translation, Speech Synthesis, etc.
Stylistics
Language Teaching
Language Pathology
Language Variation
Discourse
The following areas of study are predominantly formal:
Linguistic knowledge of a native-speaker- Phonological,
Morphological, Syntactic, Semantic etc.
Cognitive issues- nature of linguistic representation, language
processing: phonological, syntactic, orthographic, etc.
It is important to note that an investigation using one of the approaches has a lot
of scope for the development of a better theory by taking into account the
considerations of the other approach. Thus work in the computational fields of
Machine Translation and Speech Synthesis, for example, which is mainly
concerned with linguistic texts on the surface, depends on developing a formal
grammatical framework that can help make cross-linguistic comparisons and
generalizations. Similarly, work in a formal filed such as Optimality Theory has
to depend on functional principles that apply cross-linguistically. The
dependence is mutual and happening.
7. Summary
In this module, we have tried to discuss in detail the concerns in two main
approaches to language studies today. We have tried to see where they differ.
However, the main purpose of the paper has also been to show that they are
complementary rather than contending. This is so mainly because linguistic
knowledge integrates, rather than merely combines, both formal and functional
properties of language.
References
Abbney, S. P. (1987). The English Noun Phrase in its Sentential Aspects.
Cambridge, Mass.: The MIT Press.
Archangeli, Diana. (1988). Aspects of underspecification theory. Phonology 5.2:
183-207.
Baker, M. (1988). Incorporatin: A Theory of Grammatical Function Changing.
Chicago: The University of Chicago Press.
Bates, E. and MacWhinney, B. (1982). Functionalist approaches to grammar. In E.
Wanner and L. Gleitman (eds), Language Acquisition: The State of the Art.
N.Y.: CUP.
Bromberger, S. and Halle, M. (1989). Why phonology is different. Linguistic
Inquiry 20(1): 51-70.
Burton-Roberts, N., Carr, P. and Docherty, G. (eds) (2000). Phonological
Knowledge: Conceptual and Empirical Issues. Oxford: Oxford University
Press.
Bybee, Joan. (1994). A view of phonology from a cognitive and functional
perpective. Cognitive Linguistics 5: 85-305.
Carnie, A. and Mendoza-Denton, N. (2003). Functionalism is/n’t formalism: an
interactive review of Darnell et. Al. (1999). Journal of Linguistics 39, 373389.
Chomsky, N and Halle, M. (1968). Sound Pattern of English. NY: Harper and Row.
Chomsky, (1959). “On certain formal properties of grammars”. Information and
Control 2 (2): 137-167.
Chomsky, N. (1976). Problems and mysteries in the study of human of Language.
In Reflections on Language. New York: Pantheon Books.
Chomsky, N. (1986). Knowledge of Language. N.Y.: Praeger.
Chomsky, N. (1990). On formalization and formal linguistics. Natural Language
and Linguistic Theory 8: 143-147.)
Chomsky, N. (1995). The Minimalist Program. Cambridge, MA: The MIT Press.
Chomsky, N. (2001). New Horizons in the Study of Languageand Mind. Cambridge
University Press.
Comrie, B. (1981). Language Universals and Linguistic Typology. Chicago:
University of Chicago Press.
Cook, V. J. (2007). Chomsky’s Universal Grammar: An Introduction (3rd edition).
Wiley-Blackwell.
Darnell, M., Moravcsik, E., Newmeyer, F. J., Noonan, M. & Wheatley, K. (eds).
(1998). Functionalism and formalism in linguistics. Amsterdam: John
Benjamins
Dik, S. C. (1989). The Theory of Functional Grammar (Part I). Dordrecht: Foris.
Dummet, M. (1993). Language and communication. In A. George (ed.), Reflections
on Chomsky. Oxford: Basil Blackwell.
Firbas, J. (1992). Functional sentence perspective in written and spoken
communication. Cambridge: University Press.
Foley, W. A. and van Valin, R.D. (1985). Functional syntax and universal grammar.
Cambridge: Cambridge University Press.
Givon, T. (1993). English Grammar: A Function Based Introduction: Vol. I.
Philadelphia: John Benjamins Publishing Co.
Greenberg, J. (ed.) (1966). Universals of Language. Mass., Cambridge: MIT Press.
Halle, M. (1962). Phonology in a generative grammar. Word 18:54-72.
Halliday, M. A. K. (1985). An Introduction to Functional Grammar. London: Edwin
Arnold. [1st edn, 2nd edn 1994].
Jakobson, R. (1968). Selected Writings I: Phonological Studies. The Hague Mouton.
McCarthy, J. (1988). Feature geometry and dependency: a review. Phonetica 45:
84-108.
Newmeyer, F. J. (1998). Language Form and Language Function. Cambridge: MIT
Press.
Nagel, E. (1960). The Structure of Science: Problems in the Logic of Scientific
Explanation. London & Henley: Routledge & Kegan Paul.
Nichols, J. (1984). Functional Theories of Grammar. Annual Review of
Anthropology 13, 97-117.
Pandey, P. K. (2001). Grammar and ‘others’. In Anvita Abbi et al. (eds), Linguistic
Structure and Language Dynamics in South Asia. [Papers from the
Proceedings of SALAXVIII Roundtable , 1997]. Delhi: Motilal Banarasidass.
Pp. 277-290.
Patemean: T. (1990). Language in Mind and Language in Society. Oxford:
Clarendon Press.
(eds), pp 273-304.
Popper, Karl. (1963). Conjectures and Refutations: The Growth of Scientific
Knowledge. Routledge & K. Paul.
Steriade, Donca. (1995). Underspecification and markedness. In: J. Goldsmith
((ed.), The Handbook of Phonological Theory. Oxford: Blackwell. Pp. 114174.
Tomlin, R. (1986). Basic Word Order: Functional Principles. London: Croom Helm
Trubetzkoy, N. S. (1939 [1969]). Principles of Phonology, Tr. C. A. M. Baltaxe.
Berkley and Los Angeles: University of California.
###